1. Field of the Invention
The present invention relates to a maintenance system and method for a work-flow system for automatically executing processing jobs according to a work-flow defining the flow of the processing job, and more particularly to the maintenance system and method for the wide area work-flow system in which plural work-flow systems operate with relationships via a wide area network.
2. Description of the Related Art
A work-flow is a flow which defines a processing job flow, in which processing job information of who, what processing job, and in what order, is defined. A work-flow system is the system for automatically executing the processing jobs according to this work-flow. The work-flow system is a kind of groupware system which operates with a client/server system via a network in order to generally improve the productivity of the processing jobs. The historical and progressive situation for the work-flow are usually recorded in a data-base, and an administrator and worker can analyze the processing job progressive record if necessary. Further, they can change the definition of the work-flow.
A computer system capable of analyzing the work-flow by a person having no expertise, of managing a modelled work process operation, and of editing work-flow planning by changing processing job order 1, is described in a Japanese application assigned the same assignee of the present application and titled "Modelling method of the work process, tracing method of the work process model execution, and the computer system of a programming system modelling the work process" (Tokugan-Hei 6-50879, Tokkai-Hei 7-13754).
In this type of the work-flow system, the work can be linked, the processing job time is shortened, and the processing jobs operate with equal quality. Further, statistical information relating to the processing job results for each processing job can be extracted from the data-base, and there are features for analyzing any bottleneck and easily making an improvement plan for the processing job 1, by knowing which process is a time consuming process. The working process can be modified easily and the revised results are influenced immediately.
In the work-flow system based on the conventional client/server type, the management for the work-flow executed with linking between clients has been executed by one server. Thus, the processing job speed has been low since a number of clients have been sustained under only one server in order for the size of the system to become larger. So, the construction of a work-flow system using plural servers is required in order to improve the efficiency and expand the system.
However, a serious problem may arise when any work-flow system causes a system-down in the wide area work-flow system incorporating plural work-flow systems in which plural clients operate under one server and which enables each work-flow system to operate with linking via the network.
There is no problem usually because the work-flow system operates with work-flow information among servers being recorded in each data-base. However, there is a possibility of some work-flow information missing, such as historical information missing, in the data-base even if the data-base for any work-flow system server is recovered from the system-down after the system-down occurred.
The process can be executed based on the work-flow information of the data-base even if there is work-flow information missing in the data-base of each server after recovery, so the work-flow system which is closed within the server, such as one work-flow system, will not cause any operational contradiction for the work-flow system execution. On the other hand, the whole operation for the plural work-flow systems may be abandoned if flow information, such as the historical information, has some discrepancy among servers.
FIG. 1 is an explanatory view for this type of problem. Now we suppose that, in FIG. 1, the processing jobs a1 and a2 are executed in server A1, for example, and the processing jobs b1 and b2 are executed in a server B2. Then, we suppose that the processing job of a process X progresses to the processing job step of the process b1, when the processing job X is executed according to the work-flow.
At this point, work-flow historical information in the data-base of a server A1 is:
A1: the processing job a1 is executed according to the process X, and the result is input to the processing job a2; and PA1 A2: the process X is received from the processing job a1, then the processing job a2 is executed, and the result is input to the processing job b1 via the network 3. PA1 B1: the process X is received from the processing job a2 via the network and the processing job b1 is started.
On the other hand, the work-flow historical information in the data-base of a server B2 is:
At this point, we suppose that the server B2 caused a system-down and the historical information for the process was lost after the system was recovered. In this case, the problem has arisen in which the process would be shut down finally since the process X did not exist in the server B2 and subsequently the process X could not be executed, although work-flow history in the server A1 showed "the process X is completed in the server A" and the operator in the server A1 might therefore think that the processing job relating to the process X was assigned to the server B2.